‘Robotaxi’ is the buzzword of 2026. A growing number of European cities are following the US’s lead and rolling out trial fleets of autonomous taxi services; commercial launch plans are multiplying, and AI is compressing the journey from proof of concept to mass deployment faster than most transport ministers — and their regulatory frameworks — anticipated. But beneath the autonomous technology that makes the headlines lies a less glamorous, yet no less important, question: how do you run a fleet of vehicles with no one behind the wheel?
The groundwork is further down the metaphorical road than many realise. US cities already have self-driving taxi services in commercial operation, and they are generating the operational data that will in itself accelerate what is both technically and commercially possible. Across the Atlantic, a large-scale collaboration trialling self-driving, all-electric van pooling is under way in Hamburg. London now has autonomous taxis on public roads, and Zurich is joining it.
A compelling commercial logic
Removing the driver – typically the largest single operating cost in a traditional taxi service – creates an new economic model. Analysts have predicted that the global robotaxi market will grow from $1.7 billion in 2022 to $118.6 billion by 2031 – although consumer-facing price reductions are not expected until 2030, and significant upfront investment is needed to drive that shift.
A robotaxi can run around the clock, all year, limited only by maintenance windows, charging stops, and weather. Research has made it clear that autonomous vehicles are involved in fewer road accidents than human-driven ones under most conditions, although that advantage diminishes in low-light environments and on winding roads. The immeasurable safety benefits to passengers and pedestrians aside, fewer accidents mean less fleet downtime — and more revenue from each vehicle.
Scale demands more than technology
It isn’t, though, an obstacle-free road ahead, and the technical challenges are highly specific. An autonomous vehicle’s sensor array may perform flawlessly in clear daylight, but it can falter at unmarked roadworks, in heavy rain or snow, or when faced with the kind of unpredictable pedestrian behaviour an experienced driver reads almost instinctively.
Autonomous systems are still learning to interpret European roads and behaviour. The narrow, irregular layouts of even capital cities such as Rome or Paris present new obstacles to autonomous driving technologies trained on the rigid, wide-lane grid layouts of North America’s urban environments, and there can be much less separation between traffic and pedestrians. The lack of strict jaywalking laws in many European countries has presented another barrier: rather than being geofenced by known crossing points, pedestrians are often free to
cross where they please, and this has proved to be a significant hurdle for autonomous vehicles that run on algorithmic order and predictability. These are all new challenges that need to be addressed. Solvable, yes, but simple? No.
In Europe, the regulatory picture is moving quickly but unevenly, and national transport regulations still vary significantly between EU member states. The European Commission’s ADACities initiative, launched this month, signals a step up in ambition: selected cities will target fleets of 100 or more autonomous vehicles by 2030, with the goal of becoming leaders in autonomous mobility innovation. At the heart of the programme lies the potential to enhance the EU’s industrial competitiveness.
The Commission knows it is on the back foot, acknowledging that “Europe continues to lag behind global competitors” in autonomous vehicle deployment. But the regulatory gap remains real, even if the intention to close it is becoming more concrete.
Public trust is the third challenge — and perhaps the most underestimated. Studies show that just 13% of EU consumers are comfortable with a fully self-driving car, compared with a 70% trust rating in China. Notably, 82% of those surveyed said they understand AI. Europe is not unfamiliar with the technology; it is simply reluctant to hand over control.
Earning that trust requires transparency. How are vehicles monitored? What data is collected? Who bears liability when something goes wrong? These are all valid questions that the industry and regulators now have to address openly – and operational data will help to inform their answers.
Telematics: fleet intelligence at scale
Operating one robotaxi is a technology problem. Operating 1,000 of them as a commercially viable fleet is an intelligence problem. That is where telematics becomes the defining operational layer.
The automated driving system, or ADS, controls each vehicle’s actions — perception, route planning, and navigation — using onboard sensors and mapping software. Telematics sits on a different layer: a continuous data connection between every vehicle and the operator’s hub, providing fleet-wide visibility of location, vehicle health, performance diagnostics, and charging status. With no driver on board to report an unusual noise or glance at a warning light, that data feed becomes the eyes and ears of the entire network.
Battery management is where its value becomes most immediately visible. Most autonomous fleets use electric powertrains, so operators need precise, fleet-wide visibility of charging status and available range in order to deploy vehicles effectively and minimise downtime. Remote diagnostics take this a step further: robotaxis can report everything from brake wear to sensor-calibration drift, so maintenance can be scheduled around actual use rather than fixed intervals. The result is fewer breakdowns and a longer working life for each vehicle.
Fleet coordination takes place at network level. Using aggregated telematics data, the control hub can identify congestion patterns before they bite, reposition vehicles ahead of changes in demand and optimise charging schedules across an entire city. None of this affects how an individual vehicle drives — that remains the ADS’s job — but proper analysis of that data determines whether a robotaxi network operates at 90% utilisation or 60%. That’s the difference between a profitable fleet and a very expensive experiment.
A different kind of urban infrastructure
Scale the model further and the implications extend beyond mobility. The operational data that Geotab can generate from a robotaxi fleet — movement patterns, dwell times, congestion hotspots, and infrastructure stress points — can be used to give urban planners a detailed picture of how their city actually works.
Data becomes a source of insight beyond its direct operational function, providing a depth and quality of information that conventional monitoring rarely provides. The immediate commercial case rests on a fleet’s operational efficiency; the longer-term case is about what the bigger picture the data can reveal. The fleets that pull ahead will be those able to run thousands of vehicles as a single, coordinated network.
At that level, telematics stops being a supporting infrastructure and becomes a full-scale competitive advantage.